Functionally selective cannabinoid receptor signalling: Therapeutic implications and opportunities
, Giulio G. Mucciolia,b
, Emmanuel Hermansc,1,
*, Didier M. Lamberta,1
aUnite´ de Chimie Pharmaceutique et de Radiopharmacie (CMFA 7340), Louvain Drug Research Institute, Brussels, Belgium
bBioanalysis and Pharmacology of Bioactive Lipids laboratory, Unite´ d’analyse chimique et physico-chimique des me´dicaments (CHAM 7230), Louvain Drug Research Institute, Brussels, Belgium
cGroup of Neuropharmacology, Institute of Neurosciences, (UCL 54.10), Universite´ catholique de Louvain, B-1200 Brussels, Belgium
The cannabinoid receptor family currently includes two pharmacologically distinct receptors: the CB1cannabinoid recep- tor, predominantly found in the brain and other nervous tissues, and the CB2cannabinoid receptor, mainly associated with immune tissues but also expressed at a lower density in the brain.
Consistent with their widespread distribution, both cannabinoid receptors regulate a variety of central and peripheral physiological functions, including neuronal development, neuromodulatory processes, energy metabolism as well as cardiovascular, respira- tory and reproductive functions. In addition, these receptors also modulate proliferation, motility, adhesion and apoptosis of cells.
As members of the GPCR superfamily both the CB1 and CB2
cannabinoid receptors were initially reported to exert these reported biological effects by activating heterotrimeric Gi/o type G proteins . As a consequence of this preferential coupling, activation of cannabinoid receptors primarily leads to the inhibition of adenylyl cyclase and reductions in cyclic AMP accumulation in most tissues and models. In addition, both the CB1and CB2cannabinoid receptors regulate the phosphorylation and activation of different members of the family of mitogen- activated protein kinases (MAPKs), including extracellular signal- regulated kinase-1 and -2 (ERK1/2), p38 MAPK and c-Jun N- terminal kinase (JNK). In addition, CB1cannabinoid receptors can negatively couple to N- and P/Q-type voltage-operated Ca2+
channels and positively couple to A-type and inwardly rectifying K+ channels. The CB1 cannabinoid receptor may also induce elevations in intracellular Ca2+ through G protein-dependent activation of phospholipase C-
b) (Fig. 1A) (for review see ). The implication of MAPK cascades in the regulation of cell survival/death and glucose metabolism by cannabinoids, or the involvement of cannabinoid-controlled ionic currents in the regulation of neurotransmitter release are typical examples revealing that biological activities modulated by cannabinoid Biochemical Pharmacology 80 (2010) 1–12
A R T I C L E I N F O
Received 4 November 2009 Accepted 24 February 2010
Cannabinoid receptors Functional selectivity G protein-coupled receptors Cell signalling
A B S T R A C T
The CB1and CB2cannabinoid receptors are G protein-coupled receptors (GPCRs) recognized by a variety of endogenous ligands and activating multiple signalling pathways. This multiplicity of ligands and intracellular transduction mechanisms supports a complex control of physiological functions by the endocannabinoid system, but requires a ﬁnely tuned regulation of the signalling events triggered on receptor activation. Here we review the diverse signalling pathways activated by the cannabinoid receptors and discuss the mechanisms allowing for speciﬁcity in the associated functional responses triggered by endogenous or exogenous ligands. At variance with the classical concept that all agonists at a given GPCR induce a similar repertoire of downstream events in all tissues, we also summarize the experimental evidence supporting the existence of functional selectivity and protean agonism at cannabinoid receptors. By placing emphasis on the ligand- or constitutive activity-dependent speciﬁcations of receptor–G protein coupling, these concepts explain how distinct cannabinoid ligands may activate speciﬁc downstream mediators. Finally, although both the diversity and speciﬁcity in cannabinoid signalling are now establishedin vitro, few data are available fromin vivostudies. Therefore, we conclude this review by examining the experimental evidence supporting the physiological relevance of this complexity in the cannabinoid system. The ability to selectively manipulate physiological functions, through activation of deﬁned signalling cascades, will in all likelihood help in the development of efﬁcacious and safe cannabinoid-based therapeutics for a variety of indications.
ß2010 Elsevier Inc. All rights reserved.
* Corresponding author at: Group of Neuropharmacology, Institute of Neuros- ciences, (UCL 54.10), Universite´ catholique de Louvain, Av. Hippocrate 54.10, B-1200 Brussels, Belgium. Tel.: +32 2 7649339; fax: +32 2 7645460.
E-mail address:email@example.com(E. Hermans).
1Both authors contributed equally to this article.
Contents lists available atScienceDirect
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / b i o c h e m p h a r m
0006-2952/$ – see front matterß2010 Elsevier Inc. All rights reserved.
receptors are supported by complex signalling cascades. As the regulation of G protein activity constitutes the primary proximal readout of GPCR activation and underlies this multiplicity of intracellular signalling pathway regulation, it is essential to further understand the mechanisms controlling the speciﬁcity of canna- binoid receptor–G protein coupling.
While the classical view of GPCR signalling was initially to describe these receptors as simple on/off switches for the multiple intracellular cascades, recent studies have clearly indicated that they are extremely versatile signalling molecules governing
complex intracellular responses. Indeed, research over the past decade has led to growing evidence for additional, unrelated mechanisms increasing the repertoire and the complexity of GPCR- associated signalling pathways. These include: (i) the ability of the receptor to couple with distinct G proteins, (ii) the interaction with GPCR-modiﬁer proteins or with lipid raft domains to control receptor signalling or trafﬁcking, (iii) the variety of desensitization mechanisms to limit signal duration/amplitude, (iv) or the oligomerisation of GPCRs into organised novel signalling unit.
Moreover, because several experimental observations cannot be Fig. 1.Complexity at cannabinoid receptor signalling. Both CB1and CB2cannabinoid receptors are associated with Gai/o-dependent inhibition of adenylyl cyclase activity and Gbg-dependent activation of the different MAPK cascades (A). In addition, the CB1cannabinoid receptor negatively regulates voltage-gated Ca2+channels and positively regulates inwardly rectifying K+channels. Finally, the CB1receptor induces elevation of intracellular free Ca2+through Gbg-dependent activation of PLC. Cross-talks between signalling pathways are illustrated by the variety of responses requiring cannabinoid-mediated inhibition of PKA. Reduction of PKA activity is related to a reduction of gene expression through decreasing cAMP response element (CRE) activity. In addition, reduction of PKA activity leads to a decrease in constitutive inhibitory phosphorylation of c- Raf and a consecutive activation of ERK1/2. Similarly, reduction of voltage-dependent K+A channel and focal-adhesion kinase (pp125 FAK and FRNK) phosphorylations through inhibition of PKA lead to activation of these different effectors. Several of these signalling pathways are directly related to the variety of functions regulated by cannabinoid receptors. Besides, it is now demonstrated that activation of CB1cannabinoid receptors also leads to activation or Gsand Gqproteins (B). In addition the CB1
cannabinoid receptor also signals trough non-G protein partners such as the adaptor protein FAN. Preferential activation of different intracellular effectors by each G protein contributes to diversity and selectivity of responses regulated by cannabinoid receptors.
reconciled with the classical two-state receptor conformation model, several new concepts have emerged to further accommo- date the pharmacological proﬁle of several orthosteric and allosteric ligands to the experimental data, including the concepts of functional selectivity and collateral efﬁcacy, the existence of protean agonists and the possibility of probe-dependency and/or signalling pathway-dependent allosteric modulators. Consisting in a rather unique model for pharmacological studies, cannabinoid receptors are concerned by several of these properties .
Therefore, in addition to the endogenous ligand anandamide (AEA) and 2-arachidonoylglycerol (2-AG), synthetic ligands selectively interacting with CB1 or CB2 cannabinoid receptors constitute essential tools to evaluate the pharmacological proper- ties resulting from the activation of these receptors. Those that have been most often used as research tools and have contributed to unravel herein cited physiological functions are listed inFig. 2.
Focusing on several models relevant for the study of cannabinoid receptor pharmacology, this review aims at integrating the different parameters governing the diversity and the selectivity of cannabinoid receptor-mediated responses.
2. Diversity of cannabinoid-mediated signalling 2.1. Multiplicity of targeted receptors
An obvious mechanism contributing to the diversity of the intracellular signalling triggered by diverse cannabinoid ligands originates from the expression of the particular receptor subtypes themselves. Indeed, for the majority of GPCR families, multiple receptor subtypes have been identiﬁed and up to now, two cannabinoid receptors have been pharmacologically and molecu- larly characterised. Although the CB1cannabinoid receptor is the predominant receptor expressed in the central nervous system, behavioural cannabinoid-like activities have been reported in CB1
) mice following AEA administration . In addition, AEA was reported to induce guanylyl nucleotide binding in brain membranes from CB1/ animals , suggesting the existence of additional G protein-coupled AEA receptors in the
central nervous system. Similarly, an as yet unidentiﬁed non-CB1/ non-CB2endothelial cannabinoid receptor has been suggested to mediate the vasodilatator effects of some cannabinoids. While intense research, intended to identify novel cannabinoid receptors, has proposed GPR55  and GPR119  as novel cannabinoid targets, none of the reported endothelial non-CB1/non-CB2effects are associated with activation of these putative cannabinoid receptors . These data suggest that other receptors for cannabinoids and endocannabinoid ligands still remain to be identiﬁed. Likewise, cross-activity at peroxisome proliferator- activated receptors (PPARs), transient receptor potential vanilloid type 1 receptor (TRPV1), or other known receptor and non- receptor targets of cannabinoids, such as Ca2+, Na+or K+channels have been reported. Noteworthy considering the incomplete and debated characterisations of both GPR55 and GPR119 receptors in relation with the cannabinoid-mediated effects, the term canna- binoid receptor here will only refer to the CB1and CB2cannabinoid receptors.
2.2. Cross-talk between signalling pathways
Considering the selective activation of one particular receptor subtype, a second level in the complexity of cannabinoid-mediated signalling arises from the ability of a single G protein to direct the activity of unrelated intracellular effectors through either G
bgsubunits. Thus, activation of the CB1 cannabinoid receptor may simultaneously lead to the inhibition of adenylyl cyclase through G
ai/o subunits and activation of different MAPK family members through G
Further complexity is achieved by the diversity and the interconnectivity of signalling cascades regulated by the same effector. The multiplicity of cellular responses controlled by the CB1-mediated reduction of cyclic AMP accumulation and the resulting decrease in protein kinase A (PKA) activity illustrates this cross-talk (Fig. 1A). Through the classical metabolic cascade, the accumulated cyclic AMP activates PKA, which can phosphorylate the cyclic AMP response-element binding-protein (CREB) to initiate gene transcription. Therefore, the cannabinoid receptor- Fig. 2.Most frequently used cannabinoid ligands. The cannabinoid ligands that are frequently used as pharmacologic tools in cannabinoid research include either the endogenous agonists AEA and 2-AG or exogenous agonists such asD9-tetrahydrocannabinol (D9-THC), the main psychoactive cannabis component, as well as synthetic derivatives (HU 210, CP 55,940, etc.). Besides, aminoalkylindoles have been developed as more selective CB2agonists. Contrasting with this, the diarylpyrazole family includes both CB1(SR 141716A, AM 251) and CB2(SR 144528, AM630) selective antagonists/inverse agonists. While AM1241 was ﬁrst designed as a CB2selective antagonist, several studies have now proposed this ligand as a protean agonist. For a more detailed description of the pharmacologic properties of cannabinoid ligands see. Finally, more recent studies have also identiﬁed allosteric modulators of the CB1cannabinoid receptor such as Org27759 and PSNCBAM-1.
B. Bosier et al. / Biochemical Pharmacology 80 (2010) 1–12 3
mediated reduction in PKA activity may be directly correlated to an altered pattern of gene expression. In addition, inhibition of PKA has been proposed as an important pathway for ERK1/2 activation by the CB1 cannabinoid receptor in neuroblastoma cells, for activation of voltage-dependent K+ A-currents through the reduction of channel phosphorylation, and for the activation of focal-adhesion kinase (FAK) (for review see).
2.3. Multiplicity of G protein coupling
It is becoming generally accepted that a single GPCR has the ability to simultaneously activate multiple pools of related and even unrelated G proteins. Among the closely related Gi/o-type G proteins, it has been demonstrated that stimulation of the CB1
cannabinoid receptor results in the activation of various G
aosubtypes (i.e. G
ao2) in several brain regions . Similarly, co-immunoprecipitation studies have revealed the interaction of endogenously expressed CB1receptors with G
ai3isoforms in N18TG2 neuroblastoma cells . Most importantly, both the efﬁcacy and potency of the cannabinoid agonist WIN 55,212-2 vary considerably for indivi- dual G protein subtypes, suggesting that different intracellular responses are produced by the CB1cannabinoid receptor depend- ing on the preferential activation of different effectors by each G protein (Fig. 1B).
Moreover, consistent with an alternative coupling of the CB1
cannabinoid receptor with Gs proteins, cannabinoid-mediated stimulation of adenylyl cyclase has been reported following pertussis toxin (PTx, inactivating Gi/o-type G proteins) pre- treatment in several cell lines over-expressing this receptor.
Likewise, PTx pre-treatment also unmasks the coupling of the CB1cannabinoid receptor with Gq/11-type G proteins following activation by WIN 55,212-2. Arguing against the idea that multiple receptor–G protein coupling reﬂects an artiﬁcial activation of non-preferred G proteins due to receptor over- expression (promiscuity), both Gs[9,11]and Gq/11coupling has also been observed in models where cannabinoid receptors are endogenously expressed. Furthermore, it has been suggested that the successive activation of Gs- and Gi/o-type G proteins by increasing concentrations of cannabinoid agonists leads to a biphasic concentration-response proﬁle in different models, such as the biphasic regulation of voltage-gated Ca2+currents [13,14], or to a biphasic regulation of GABA release . This may provide a mechanism preventing excessive cell excitability upon robust receptor activation. In a further demonstration of the physiological relevance of these alternative couplings a recent study presented evidence that cannabinoid tolerance induced by WIN 55,212-2 was associated with a molecular switch from Gi/oto Gscoupling in striatum. Therefore, one may propose that the physiological activation of different unrelated G protein species provides a complex mechanism allowing for both the ﬁne-tuning and the adaptation of diverse functional responses elicited by CB1 cannabinoid receptor activation.
Finally, the recruitment of intracellular effectors in a G protein-independent manner is also thought to play a pivotal role in the temporal resolution of cannabinoid-mediated responses. Hence, transient CB1 receptor activation initiates sphingomyelin breakdown and ceramide accumulation through functional coupling with the adaptor protein FAN, while a sustained stimulation promotes G protein-dependent de novo ceramide synthesis through activation of serine palmitoyl- transferase activity , and these respective acute and sustained accumulations of ceramide induced by cannabinoids were related to the regulation of metabolic functions and cell survival/death decision.
3. Speciﬁcity of responses mediated by cannabinoid receptor activation
3.1. Control of cannabinoid-mediated signalling
As most GPCRs have been shown to interact with different G proteins, one might expect a consequent lack of speciﬁcity in downstream signal transduction mechanisms. However, it is likely that each GPCR displays its own and distinctive repertoire of coupling with G proteins. Thus, the CB1cannabinoid receptor has been shown preferentially to couple to a deﬁned subset of G
subunits, distinct from those speciﬁcally activated by other GPCRs. Indeed, several factors, both extra- and intracellular, appear to promote the speciﬁcity of cannabinoid signalling and conﬁne cannabinoid responses in a spatial and temporal fashion (Table 1).
While these mechanisms emphasize the versatility of GPCR signalling they also constitute a crucial and synchronized machinery to limit signal scattering and constrain cannabinoid responses.
With respect to the intracellular components of signal transduction, several studies have reported that CB1cannabinoid receptor-mediated signalling shares a common pool of G proteins with adrenergic, somatostatin, insulin and IGF-1 receptors, while this receptor activates a separate pool of G proteins than those regulated by purinergic and GABAergic receptors. This undoubtedly points out the expression and the availability of distinct G protein sub-populations as crucial determinants of signalling speciﬁcity. Furthermore, recombinant opioid and cannabinoid receptors were shown to operate through the same pool of G proteins only in transfected cells, whereas in cells endogenously co-expressing these receptors, these receptor subtypes signal through distinct pools of G proteins, suggesting that cells possess subcellular organisation allowing for further speciﬁcity of receptor–G protein interactions.
Nevertheless, the complexity of GPCR signalling is not only a consequence of the multiplicity of G protein coupling, but also results from the propensity of the receptor machinery to function as a deﬁned complex involving a variety of cellular protein partners which preorganise GPCR signalling. Indeed, interactions of cannabinoid receptors with several ‘‘GPCR-modiﬁer proteins’’
participating in the ﬁne-tuning of signalling or trafﬁcking of the receptor have recently been reported (see examples inTable 1).
Besides, it has now been demonstrated that CB1 cannabinoid receptors exist as multimers (either homo- or heteromers) resulting in receptor combinations displaying altered binding properties and that signal and trafﬁc differently from the GPCR monomer. Finally, considering GPCRs as well-deﬁned, but dynamic proteins, alterations in the receptor structure would likely affect cannabinoid signalling properties. As summarized inTable 1, both post-transcriptional and post-translational modiﬁcations of the CB1cannabinoid receptor further contribute to alterations in its pharmacological properties.
The speciﬁc insertion of GPCRs within different membrane compartments, such as lipid rafts and non-raft domains, is also likely to determine signalling selectivity. Accordingly, a recent study has reported the inﬂuence of cell compartmentalization on CB1cannabinoid receptor binding and signalling. In parallel, both the synthesis and uptake of endocannabinoids was shown to be associated with lipid rafts, reinforcing the relevance of membrane microdomains for the endocannabinoid system.
Finally, a striking characteristic of cannabinoid signalling adaptation is the variation in the magnitude and kinetics of CB1
cannabinoid receptor desensitization and down-regulation docu- mented in different brain regions. This supports the concept that prolonged exposure to cannabinoids may result in different adaptation proﬁlesin vivo. As an example, regional differences in
ERK cascade involvement in the adaptive processes of the CB1
cannabinoid receptor have been reported and are suggested to account for the development of tolerant and/or addicted states to cannabinoids. As the mechanisms involved in CB1cannabinoid receptor desensitization and down-regulation are essential to limit signal duration as well as in determining signal quality, these regional differences may provide interesting insights into the mechanisms of CB1receptor signalling in different brain regions.
3.2. Cell-based selectivity of cannabinoid-mediated signalling
As already suggested, the speciﬁcity of cannabinoid receptor- mediated responses largely relies on the inclusion of the receptor in deﬁned signalling complexes. Therefore, variation in the expression level of the different G proteins and others signalling partners, as well as their presence in the vicinity of the receptor and their organisation into signalling complexes is expected to support cell-based selectivity and tissue differences in the pattern of cannabinoid-mediated responses. In this regard it is interesting to note that the CB1cannabinoid receptor agonist WIN 55,212-2 has been shown to differ in its ability to stimulate guanylyl nucleotide binding, and to differentially modulate the number of activated G proteins per unit of occupied receptor depending on the brain region . Further supporting the concept that
intracellular signalling may vary across different neuronal environments, weak or an absence of inhibition of adenylyl cyclase was detected in certain brain structures in which the CB1
receptor was nevertheless shown to activate G proteins. As the coupling to G proteins is reported to be Na+-sensitive, the regional differences in the sensitivity of cannabinoid-mediated adenylyl cyclase inhibition to the Na+concentration possibly supports the existence of brain regional speciﬁcity in the subtypes of G proteins activated by the CB1cannabinoid receptor.
Contrasting with the brain regional differences in adenylyl cyclase regulation, other studies have only reported modest differences in the activation of G protein subtypes by WIN 55,212-2 . This corroborates the widespread concept that achieving cell-based selectivity not only requires the efﬁcient interaction of GPCRs with G proteins, but also depends on the appropriate interactions between G proteins and downstream effectors. Thus, a given ligand may act as a partial agonist in well- coupled tissues but as an antagonist in tissues were receptor coupling and/or G protein-dependent effector activation is less well coupled.
Closely related to this, it has been proposed that the density of receptors plays a major role in governing the amplitude of response to partial and full cannabinoid agonists . Indeed, increasing receptor expression, relatively to the level of G proteins Table 1
Factors putatively inﬂuencing CB1and CB2cannabinoid receptor signalling.
Interaction levels Interfering factors Effects on cannabinoid signalling References
Extracellular Presence of co-activators for other GPCRs
Convergence of dopamine and cannabinoid signalling pathways 
Alteration of CB1receptor coupling upon D2 receptor coactivation 
Attenuation of CB1signalling bymopioid receptor ligands 
Reduction of CB1receptor-mediated G protein activation by GABA B receptor antagonist
Regulation of Gi/o-dependent CB1receptor signalling through coactivation of A2Areceptor
Modulation of cannabinoid agonist binding by serotonin 
Co-stimulants for intracellular effectors
Potentiation or alteration of CB1-mediated signallings in the presence of forskolin
Intracellular Availability of different G proteins
Dopamine D2receptor-mediated sequestration of Gi/oproteins [15,77]
Expression of different effector isoforms
Inﬂuence of adenylyl cyclase isoforms on the global cAMP outcome regulated by CB1receptor
Expression of GPCR interacting proteins
Various effects of CRIP1a on various CB1signalling pathways regulated by different ligands
Involvement of CB1receptor/GASP interaction in spatial and temporal receptor trafﬁcking
Interaction of CB1receptor with M6a scaffolding protein 
Regulation of CB1trafﬁcking to the lysosome by AP-3 
Role of molecular chaperone Hsp90 to ensure CB2receptor proper coupling and signalling pathways
Receptor Splice variants Expression in different tissues of CB1receptor splice variants possessing unique pharmacological proﬁle
Tissue-selective expression of CB2receptor isoforms 
Phosphorylation Phosphorylation of CB1receptor as a mechanism to regulate G protein coupling and desensitization
Nitrosylation Reduction of CB1signalling through S-nitrosylation of the receptor 
Dimerisation Expression of CB1homodimers in several brain regions 
Allosteric interactions between opioid and CB1receptors [71,89]
Heterodimer/heteromer formation containing D2and CB1receptors [90,91]
Heterodimer formation between orexin-1 and CB1receptors and between A2Aand CB1receptors
Compartmentalization Expression of CB1receptor in intracellular vesicles
Gai-dependent regulation of ERK activity by late endosomal-lysosomal vesicle CB1receptors
Membrane microdomains Control of CB1receptor binding, signalling or trafﬁcking by lipid rafts [93,94]
Reduction of CB1receptor binding and signalling by cholesterol enrichment of cell membranes
Non-exhaustive list of factors inﬂuencing the functional responses associated with cannabinoid receptor signallings.
B. Bosier et al. / Biochemical Pharmacology 80 (2010) 1–12 5
that can be maximally activated, preferentially enhances the efﬁcacy of partial agonists . Consistent with this, both anatomical and signalling pathway-dependent differences in receptor reserve have been proposed to explain variations in agonist efﬁcacies and potencies for different signal transduction mechanisms across brain regions . Alternatively, it was suggested that a reduction in CB1 cannabinoid receptor density (obtained in transgenic heterozygote CB1+/ mice) leads to an increased G protein coupling efﬁciency . Providing rather convincing indications that brain regional speciﬁc responses could result from differences in the receptor density, this study also showed that the consequence of decreasing CB1 cannabinoid receptor density on coupling efﬁciency varied among the different areas studied.
Beside these cell-dependent constraints, the speciﬁcity of cannabinoid receptor-mediated responses also relies on the extracellular environment of the targeted cells. Undoubtedly, on-demand release, rapid degradation and retrograde versus anterograde release of endocannabinoids within a highly localized space provides an effective means to control cannabinoid responses. On the other hand, evidence suggests that concomitant release of other mediators or GPCR ligands in the proximity of the cannabinoid receptor may subsequently modify basal cell tone, or interfere with cannabinoid signalling pathways. As an example, convergence of dopamine D2 and CB1 receptor-mediated signal transduction may contribute to the considerable inﬂuence of dopaminergic ligands on CB1-induced behavioural responses observed in both rodents and humans[32,33].
4. In vitrohints for further complexity at cannabinoid receptors 4.1. Beyond the two-state conformational model
Considering classical two-state receptor theory, the efﬁcacy of ligands reﬂects the predominant stabilization of a deﬁned receptor conformation, among the active and inactive conformations, and between which the receptor oscillates (active (R0) for agonists; and inactive (R) for inverse agonists; no modiﬁcation of the R–R0 equilibrium for antagonists). Thus, the ligand proﬁle, also deﬁned as its intrinsic efﬁcacy, can be viewed as the ratio of the afﬁnities for R–R0 conformations. Nevertheless, due to experimental limitations, the quantiﬁcation of the proportion of receptors in each conformation is hardly accessible and classiﬁcation as either full/partial agonist, full/partial inverse agonist or antagonist rather relies on the evaluation of functional responses mediated by receptor ligands.
Classically, only the predominant signalling pathway or biological response is considered when deﬁning the efﬁcacy of a wide variety of ligands. Therefore, reduction in cyclic AMP levels is commonly assessed to determine endogenous and exogenous cannabinoid ligand efﬁcacy. The development of techniques measuring the exchange of guanylyl nucleotides on receptor activation (e.g. [35S]-GTP
gS binding assays) has given access to more direct quantiﬁcation of receptor/G protein coupling, which is frequently used to estimate cannabinoid ligand efﬁcacy. However, the complexity of certain responses associated with several cannabinoid drugs, deﬁned as CB1 or CB2 agonists, using the [35S]-GTP
gS binding assay (see examples inTable 2) can be best explained by considering a model in which the receptor can independently couple to different signalling pathways. Hence, even though both HU 210 and CP 55,940 are established as full CB1
and CB2receptor agonists in [35S]-GTP
gS binding assays[29,34], they have been described as partial agonists when examining their capacity to induce intracellular Ca2+mobilisation . Further- more, the endogenous cannabinoid AEA, which is generally reported as a partial agonist at both CB1 and CB2 cannabinoid
receptors , has also been found to display full agonist properties in different assays, further questioning the physiolo- gical signiﬁcance of such discrepancies between signalling path- way-related efﬁcacy and/or potency. While the different effects presented inTable 2 were all prevented using CB1 cannabinoid receptor antagonists, a major criticism to the full agonist proper- ties of AEA arises because of evidence for a non-CB1 receptor component of the AEA-mediated response, which is thought to contribute to the inhibition of N-type Ca2+currents.
Therefore, the assumption that the interaction of an agonist with its receptor leads to equivalent efﬁcacy towards the different transduction pathways is inadequate to explain these unexpected cannabinoid-mediated responses. While the reported non-canna- binoid receptor-mediated responses  should be taken into account when estimating the efﬁcacy and potency of a ligand, the picture of an agonist interacting with a unique receptor-active conformation to promote all receptor functions is overly simplistic.
Rather, evidence suggests that GPCRs exist in multiple active receptor conformations displaying distinct abilities to regulate individual signalling pathways. Based on their respective afﬁnities for each receptor conformation, it is predicted that different ligands could ultimately produce distinct functional effects by inducing selective enrichment of individual active receptor conformations. While the studies presented in Table 2 do not constitute irrevocable evidence for this mechanism, generally referred as agonist-directed trafﬁcking of GPCR signalling or functional selectivity, allows the reconciliation between efﬁcacy and potency discrepancies reported among the different signalling pathways. Furthermore, considering the wide variety of mechan- isms inﬂuencing GPCR signalling (see Table 1), which include interaction with GPCR-modiﬁer proteins, the insertion into lipid rafts, or the presence of receptor reserve, one cannot exclude a contribution of these mechanisms to the selectivity of the different cannabinoid ligands among signalling cascades.
On the other hand, with regard to receptor regulation and trafﬁcking processes, one could suggest that distinct agonist- activated conformations of a GPCR could undergo independent molecular desensitization . Indeed, challenging the linear concept that by increasing the number of receptors in an active conformation, more effective agonists lead to more pronounced desensitization and internalization, Luk et al. have identiﬁed a highly potent and effective cannabinoid agonist that induces slow receptor desensitization . This unrelated regulation could be best explained by considering the existence of different desensitization prone receptor conformations stabilized by distinct agonists.
4.2. Constitutive activity and protean agonism
The extension of the two-state conformational model to a multiple active-state conformational model certainly helps to explain the above mentioned discrepancies in ligand efﬁcacies.
However, considering the functional properties of ligands as a continuum from inverse agonism to full agonism, variations in the constitutive activity of the receptor adds further complexity to the outcomes of interaction between cannabinoid ligands and their receptors . Indeed, it has been suggested that any factor increasing the constitutive activity of the CB1cannabinoid receptor should produce an increase in inverse agonist efﬁcacy and a decrease in agonist efﬁcacy, as was demonstrated for a C-terminally truncated CB1cannabinoid receptor (CB1(
D417)). Thus, in neurons expressing CB1(
D417), SR 141716A produced a greater enhancement and WIN 55,212-2 a smaller inhibition of Ca2+signalling compared to neurons expressing the wild-type receptor. Similarly, different canna- binoid agonists showed a restricted proﬁle of G protein activation when tested on a mutated CB1 cannabinoid receptor displaying enhanced basal G protein activation.
Examples of discrepancies among pharmacological responses reported at CB1and CB2cannabinoid receptors.
Ligands Measured responses Efﬁcacy Models References
Pharmacological responses reported for the CB1cannabinoid receptor
AEA [35S]-GTPgS binding Partial agonist (around 75% of max response to
WIN 55,212-2, HU 210 or CP 55,940)
Rat cerebella membranes 
Inhibition of forskolin-stimulated cAMP formation Full agonist compared to CP 55,940 Human neocortical synaptosomes 
Inhibition of N-and P/Q- types channels-mediated Ca2+currents
Full agonist compared to CP 55,940 and WIN 55,212-2
Rat hippocampal neurons 
Inhibition of Q-type channels-mediated Ca2+currents, and increase in GIRK-mediated K+currents
Full agonist compared to WIN55,212-2 AT20 cells expressing rat CB1receptor 
Inhibition of cell proliferation Full agonist compared to HU 210 and 2-AG EFM-19 human breast cancer cell line 
Inhibition of electrically evoked acetylcholine release Full agonist compared to WIN 55,212-2 Hippocampal brain slice 
Artery relaxation Full agonist compared to WIN 55,212-2 Bovine ophthalmic artery 
CP 55,940 [35S]-GTPgS binding Full agonist Rat cerebellum/brain membranes
Increase in the intracellular Ca2+concentration Partial agonist (around 45% of max response to WIN 55,212-2
Rat hippocampal neurons 
Increase in the intracellular Ca2+concentration Partial agonist (around 40% of max response to 2-AG) NG108-15 cells 
Increase in GIRK1/4-mediated K+currents Partial agonist (around 30% of max response to WIN 55,212-2
Xenopus oocytes expressing CB1h 
HU 210 [35S]-GTPgS binding Full agonist Rat cerebellum membranes CHO-CB1h [29,34]
Increase in intracellular Ca2+concentration Partial agonist (around 40% of max response to 2-AG) NG108-15 cells 
Inhibition of electrically evoked contraction Partial agonist (around 25% of max response to WIN 55212-2
Rat isolated urinary bladder 
Pharmacological responses reported for the CB2cannabinoid receptor 2-AG
[35S]-GTPgS binding Full agonist
EC50: 38.9 nM and 1.1 nM for 2-AG and HU 210
Inhibition of forskolin-stimulated cAMP formation Low potency compared to HU 210 (EC50 1.3mM and 1.6 nM for 2-AG and HU 210)
AEA [35S]-GTPgS binding Partial agonist (around 35% of max response
to HU 210)
EC50121 nM and 1.1 nM for AEA and HU 210a
Inhibition of forskolin-stimulated cAMP formation Nearly ineffective, very low potencyacompared to HU 210
EC50>30mM and 1.6 nM for AEA and HU 210a
IL8 production Inactive compared to CP 55,940 Human promyelocytic leukemia
CP 55,940 [35S]-GTPgS binding Full agonist CHO-CB2h 
Release of IL2 Inactive compared to WIN 55,212-2 Human blood mononuclear cells 
In this table, functional responses observed in [35S]-GTPgS binding assays (presented in bold) were used to classify CB1and the CB2cannabinoid receptor ligands into full or partial agonists. While in the majority of functional studies (not reported in this table) these ligands behave similarly to what is observed in the nucleotide binding assays, this table reports studies in which a different pharmacological response was observed.
a Differences in relative potency to HU 210 were reported: relative potency = EC50 tested compound/EC50 HU 210. [35S]-GTPgS relative potency: 35 and 110 for 2-AG and AEA, respectively. cAMP accumulation relative potency:
812 and>1875 for 2-AG and AEA, respectively.
In biological systems, the constitutive activity of a single GPCR, which is determined by the equilibrium between active and inactive conformations, is governed by several extrinsic cell- dependent factors (i.e. cell environment, GPCR-modiﬁer proteins, post-transcriptional and/or post-translational modiﬁcation, etc.).
Therefore, one may predict that ligands with modest intrinsic efﬁcacy might induce opposite functional responses (either positive agonism in system displaying low constitutive activity, or inverse agonism in system displaying high constitutive activity).
This concept, proposing ligands with modest intrinsic efﬁcacy to exhibit partial agonism or inverse agonism depending on receptor constitutive activity is termed ‘‘protean agonism’’. Accordingly, based on differences in the regulation of cyclic AMP accumulation in recombinant versus native receptor-expressing systems, AM1241 has been proposed to act as a protean agonist at the CB2cannabinoid receptor. While large differences in receptor expression level could interfere with the responses mediated by AM1241 in these models, an independent study attempted to demonstrate the protean nature of this agonist using a different strategy. By manipulating adenylyl cyclase activity with forskolin, Yao et al. observed different efﬁcacies for AM1241 . The concept that selectivity of the response depends on both the intrinsic efﬁcacy of the ligand and on the basal tone of the targeted system was corroborated by showing that the pharmacological proﬁle of AM1241 may be switched from antagonist to agonist by suppressing CB2cannabinoid receptor constitutive activity.
5. Pharmacological manipulation of CB1receptor coupling 5.1. Agonist-supported modulation of signalling speciﬁcity
Evidence showing that distinct agonists can differentially regulate signalling pathways through selective activation of G proteins has accumulated for majority number of GPCRs. From a pharmacological point of view, this implies that selectivity for speciﬁc biological responses not only arises through the selective targeting of a single GPCR subtype, but also relies on the selection of a particular signalling pathway by the activated receptor.
Constituting some of the earliest evidence for ligand-directed receptor–G protein coupling, Na+ ions and guanine nucleotides were reported to inﬂuence the efﬁcacies of CB1 cannabinoid receptor–G protein coupling in a ligand-dependent manner.
Other studies have reported on agonist-speciﬁc differences in potency and/or efﬁcacy towards activation of different G protein species on CB1receptor activation. Thus, Glass and Northup have demonstrated that cannabinoid agonists differ in their ability to activate individual Gi- and Go-type G proteins. Conﬁrming the concept of functional selectivity at the CB1cannabinoid receptor, co-immunoprecipitation studies have revealed that through interaction with this receptor WIN 55,212-2 activates all subtypes of Giproteins (Gi1, Gi2, Gi3) while the unrelated cannabinoid ligand, desacetyllevonantradol, activates Gi1 and Gi2 while acting as an inverse agonist with respect to Gi3coupling. Extending this concept to other types of G protein, agonist-selective regulation of Gsand Gqcoupling to the CB1cannabinoid receptor have been reported. Further supporting this model where different conformations of the receptor are required for selective proﬁles of G protein activation, a mutagenesis study has recently provided a molecular basis for the differential coupling observed with chemically distinct agonists . In addition, Georgieva et al.
found that the structurally unrelated cannabinoid ligands CP 55,940 and WIN 55,212-2, may stabilize distinct active conforma- tions of the CB1 cannabinoid receptor that exhibit differential abilities to interact with- and activate the Gi1protein.
Finally, the question of how such agonist-selective coupling of the CB1 cannabinoid receptor may translate into differential
activation of cell signalling and function has been further investigated in biochemical studies directly comparing agonist efﬁcacies and potencies at different levels of the downstream signalling cascades. By comparing the transcriptional regulation achieved by HU 210 and CP 55,940, Bosier et al. reported on unpredicted different proﬁles of these two ligands, which are commonly viewed as reference agonists at the CB1cannabinoid receptor. Indeed, these ligands were found to different effects on the expression of tyrosine hydroxylase in a model of neuroblas- toma cells: while both ligands inhibited CRE-dependent gene transcription, CP 55,940 acted as an antagonist on the AP-1- dependent gene transcription, whereas HU 210 behaved as a full agonist . In addition, it was demonstrated that among the different MAPK family members, HU 210 was more efﬁcacious in increasing ERK1/2 phosphorylation, while CP 55,940 displayed a higher efﬁcacy to activate JNK . While agonist-selective coupling was not speciﬁcally investigated in these studies, it was suggested that these differences could support the agonist- selective regulation of tyrosine hydroxylase gene expression, emphasizing the putative pharmacological consequences of functional selectivity. Similarly, questioning the involvement of controlled signal diversiﬁcation in physiology, a recent study identiﬁed analogues of hemopressin as novel endogenous peptide agonists of the CB1 cannabinoid receptor which activate signal transduction pathways distinct from those activated by endocan- nabinoids and HU 210.
Surprisingly, while agonist-selective coupling or differential receptor conformations have not been described yet for the CB2
cannabinoid receptor, dissimilarities in the potencies of agonists tested for independent intracellular cascades have been reported, supporting functional selectivity at this subtype. Thus 2-AG was more potent in activating ERK than in inhibiting adenylyl cyclase and regulating Ca2+ transients, whereas noladin ether and CP 55,940 most potently inhibited adenylyl cyclase. Furthermore, in myeloid precursor cells, CP 55,940 was reported to alter neutrophilic differentiation, whereas 2-AG was an efﬁcient stimulator of cell migration , highlighting the potential relevance of identifying functionally selective agonists.
In summary, this compilation of experimental observations demonstrates that CB1or CB2cannabinoid receptors may support complex, but selective pharmacological regulation depending on the particular proﬁle of the ligand used. Therefore, one may expect unique patterns of functional interactions arising both in cell line (or tissue)- and ligand-dependent manners.
5.2. Allosteric modulation of selectivity
Due to their ability to interact independently with a remote binding site on the receptor, allosteric modulators display an array of effects either dependent or independent of the occupancy of the orthosteric binding site. Thus, allosteric ligands alter orthosteric ligand afﬁnity or efﬁcacy through diverse allosteric ligand stabilized conformational changes. In addition, by directly inﬂuencing the coupling of the receptor, some allosteric mod- ulators possess their own efﬁcacy, thereby directly regulating signalling pathways, without requiring activation of the receptor by an orthosteric ligand. In this view, the receptor/allosteric modulator complex might be considered as a ‘‘new GPCR’’ entity displaying altered response selectivity.
Among the existing strategies allowing regulation of CB1
receptor-mediated responses, allosteric modulators of the CB1
cannabinoid receptor provide an innovative way to modulate signalling [57–59]. These compounds, including Org 27759, display markedly divergent effects on orthosteric ligand afﬁnity versus efﬁcacy: for example, they might be allosteric enhancers of agonist binding and allosteric inhibitors of agonist signalling
efﬁcacy. Furthermore, these allosteric modulators display ligand- dependent effects, thereby inhibiting the binding of an inverse agonist. Based on the two-state conformational model, it is proposed that Org 27759 traps the receptor in a high afﬁnity, coupled state explaining differential inﬂuences on agonist and antagonist afﬁnities. Recently, a novel allosteric modulator, PSNCBAM-1, displaying a similar pharmacological proﬁle has been identiﬁed. Importantly, the use of this novel compound inin vivostudies focusing on the control of food intake and body weight sheds new light on the potential relevance of CB1receptor allosteric modulators in pharmacology . Since allosteric modulation is foreseen as a novel advantageous strategy in the therapeutic exploitation of GPCRs future research will undoubt- edly address the question of whether allosteric modulators of CB1
cannabinoid receptors are advantageous over orthosteric ligands in particular pathophysiological scenarios.
6. Physiological and pharmacological implications of functional selectivity at cannabinoid receptors
As we have reported here, evidence has accumulated for a variety of mechanisms inﬂuencing CB1 cannabinoid receptor signalling which allow for both the diversiﬁcation and the speciﬁcation of cannabinoid-mediated responses. Nevertheless, the physiological implications of such complexity in the regulation of receptor signalling and the relevance of agonist-selective responses remain elusive. Indeed, pharmacokinetic differences aside, the ﬁnal in vivo outcome of cannabinoid exposure simultaneously depends on the functional properties of the ligand and on the pre-existing organisation of receptor signalling complexes in those tissues reached by the ligand (Fig. 3). The most convincing examples of functional selectivity are obtained from studies where different ligands exhibit different rank-order of
potencies or efﬁcacies with respect to independent intracellular signal transduction pathways within the same cell. Even though thein vivodissection of these intricate phenomena is to date hardly conceivable, studies have already reported on selective cannabi- noid responses depending either on the animal model or on the ligand used.
Historically, the characterisation of cannabinoid ligands has often relied on four behavioural tests (the tetrad, including anti- nociception, hypoactivity, hypothermia and catalepsy), affording the ﬁrst indications of differences between in vivo activities of cannabinoids. As a correlation was demonstrated between potencies in the behavioural assays and binding afﬁnity at the CB1cannabinoid receptor, most of the cannabinoid ligands were expected to show equivalent efﬁcacies in the four facets of the tetrad test. Nevertheless, discrepancies in the rank-order of efﬁcacies or potencies have frequently been reported. As an example,
D9-THC and WIN 55,212-2 are more potent in inducing hypolocomotion than catalepsy or hypothermia, whereas several WIN 55,212-2-derived ligands are more potent in inducing catalepsy or hypothermia . Although the selectivity of CB1
receptor interactions was not assessed, these studies suggest that different cannabinoid agonists may produce distinct combinations of physiological responses.
Underlining the critical outcomes possibly arising from ligands displaying distinct pharmacological proﬁles, disparities among the potencies of AEA and
D9-THC in inducing anti-nociception and hypolocomotion have been reported. Corroborating with the contribution of agonist-selective signalling in the regulation of disparate in vivo functions, differences both in the control of nigrostriatal neurons activity  and in the CB1-dependent pattern of interactions with opioid system  have been described. Furthermore, a study of Glass and Northup showed that AEA and
D9-THC are partial agonists towards the Go-type G
Fig. 3.Different levels of CB1signalling modulation. Several mechanisms, interfering at different levels of the signalling cascade are now reported to inﬂuence the coupling of CB1cannabinoid receptor with different G proteins. Due to a ﬁne-control by both intra- and extracellular factors, the signalling cascades regulated by CB1cannabinoid receptor are cell-speciﬁcally preorganised. Indeed several interactions with cellular proteins, which are in general conﬁned to deﬁned membrane microdomains, serve to organise the different partners of signalling cascade, modulate coupling efﬁciency and control receptor trafﬁcking. Therefore complex machinery exists in the environment of the receptor to precisely govern its cellular responses (cell-based selectivity). Nevertheless, pharmacologists currently dispose of several tools to modulate these preorganised signallings. Obviously the ligand proﬁle is crucial in dictating the responses. Besides the classically described agonists, inverse agonists and antagonists, ligands possessing more complex proﬁles (e.g. functionally selective agonists, protean agonists and allosteric modulator) are shown to independently manipulate the diverse cellular responses. In addition, because different co-activator or other GPCR ligands interact with cannabinoid signalling, approaches based on compound coadministration could also be developed.
B. Bosier et al. / Biochemical Pharmacology 80 (2010) 1–12 9
protein whereas only AEA is a full agonist towards Gi-type G protein. Even more convincingly, the use of selective FAAH inhibitors revealed that AEA induces anti-nociception without causing hypothermia or hypomobility, while the use of a selective monoacylglycerol lipase inhibitor revealed that 2-AG modulates all the behavioural processes classically attributed to cannabinoids , suggesting a segregation of endocannabinoid signalling in vivo. While throughout these different studies, the functional selectivity could contribute to the manifestation of such selective behaviours, one cannot exclude that differences in the metabolism, tissue distribution or targetting of receptors may also inﬂuence these physiological responses, especially when comparing data obtained with synthetic ligands and endocannabinoids.
Questioning the inﬂuence of receptor constitutive activity on the functional outcome measured in different systems, some CB1
cannabinoid receptor partial agonists have been shown to exacerbate formalin-induced hyperalgesia in spite of exhibiting anti-nociceptive properties. Similarly, theR-enantiomer of AM1241 was reported to induce CB2 receptor-dependent anti- nociception in rats despite its inverse agonist proﬁle determined using cyclic AMP measurements in recombinant cell lines expressing the rat CB2cannabinoid receptor. While strongly suggestive of protean agonism, these authors suggested the involvement of an as yet undeﬁned mechanism. Even though direct comparisons of constitutive activity of the receptor in particular cell lines and animals are lacking to conﬁrm the hypothesis of protean agonism, these studies clearly demonstrate that the functional properties elicited by different ligands are strongly dependent on the model used.
In accordance with the critical inﬂuence of the model on the physiological outcomes of cannabinoids, the CB1receptor-selective antagonist/inverse agonist SR 141716A has been shown to alter motor behaviours in different animal models of Parkinson’s disease, but not in control animals. Interestingly, behavioural changes were associated with differences in CB1 cannabinoid receptor expression and signalling. Together, these ﬁndings suggest that altering the cellular environment by modulating G protein/effector expression and/or basal activity by concomitantly activating diverse intracellular cascades strongly interferes with cannabinoid signalling (Fig. 3).
7. Concluding remarks
Since the discovery of the cannabinoid receptors and their endogenous ligands, the endocannabinoid system has been regularly regarded as a putative target for the treatment of several diseases, including neurodegenerative diseases (Alzheimer’s dis- ease, Parkinson’s disease, Huntington’s disease, etc.), cancer, neuropathic and inﬂammatory pain, obesity, etc. Nevertheless the potential clinical uses of cannabinoids remain strongly limited by the unacceptable adverse effects of cannabis including its psychotropic action.
While remarkable advances in the development of highly selective agonists have emerged during this last decade, present studies indicate that speciﬁcity in cannabinoid-mediated functions is not only achieved by the pharmacological proﬁle of the ligand used but also depends on cell/model-related parameters. There- fore, the ability to selectively manipulate different physiological functions by targeting either a subpopulation of receptors or a deﬁned associated signalling cascade will certainly constitute the basis of novel and promising therapeutic approaches. Along this line, the observation that some
D9-THC derivatives are equally potent to
D9-THC in inducing anti-nociception, while being 30–40 fold less potent in inducing hypothermia, hypoactivity or catalepsy is encouraging. Certainly, a more thorough characterisation of the versatile nature of cannabinoid signalling is essential to
optimize the development of cannabinoid ligands as therapeuti- cally safe drugs.
This study was supported by grants from the National Fund for Scientiﬁc Research (FNRS) and from UPSA (Institut de la douleur).
BB and EH are senior research assistant EH and Research Director of the F.N.R.S., respectively.
 Howlett AC. Cannabinoid receptor signaling. Handb Exp Pharmacol 2005;53–79.
 Hudson BD, Hebert TE, Kelly ME. Ligand and heterodimer directed signaling of the CB1 cannabinoid receptor. Mol Pharmacol 2009;77:1–9.
 Di Marzo V, Breivogel CS, Tao Q, Bridgen DT, Razdan RK, Zimmer AM, et al.
Levels, metabolism, and pharmacological activity of anandamide in CB(1) cannabinoid receptor knockout mice: evidence for non-CB(1), non-CB(2) receptor-mediated actions of anandamide in mouse brain. J Neurochem 2000;75:2434–44.
 Ryberg E, Larsson N, Sjogren S, Hjorth S, Hermansson NO, Leonova J, et al. The orphan receptor GPR55 is a novel cannabinoid receptor. Br J Pharmacol 2007;152:1092–101.
 Overton HA, Babbs AJ, Doel SM, Fyfe MC, Gardner LS, Grifﬁn G, et al. Deor- phanization of a G protein-coupled receptor for oleoylethanolamide and its use in the discovery of small-molecule hypophagic agents. Cell Metab 2006;3:167–75.
 Hiley CR, Kaup SS. GPR55 and the vascular receptors for cannabinoids. Br J Pharmacol 2007;152:559–61.
 Prather PL, Martin NA, Breivogel CS, Childers SR. Activation of cannabinoid receptors in rat brain by WIN 55212-2 produces coupling to multiple G protein alpha-subunits with different potencies. Mol Pharmacol 2000;57:1000–10.
 Mukhopadhyay S, Howlett AC. CB1 receptor–G protein association. Subtype selectivity is determined by distinct intracellular domains. Eur J Biochem 2001;268:499–505.
 Glass M, Felder CC. Concurrent stimulation of cannabinoid CB1 and dopamine D2 receptors augments cAMP accumulation in striatal neurons: evidence for a Gs linkage to the CB1 receptor. J Neurosci 1997;17:5327–33.
 Lauckner JE, Hille B, Mackie K. The cannabinoid agonist WIN55,212-2 increases intracellular calcium via CB1 receptor coupling to Gq/11 G proteins.
Proc Natl Acad Sci U S A 2005;102:19144–9.
 Bash R, Rubovitch V, Gafni M, Sarne Y. The stimulatory effect of cannabinoids on calcium uptake is mediated by Gs GTP-binding proteins and cAMP formation. Neurosignals 2003;12:39–44.
 McIntosh BT, Hudson B, Yegorova S, Jollimore CA, Kelly ME. Agonist-depen- dent cannabinoid receptor signalling in human trabecular meshwork cells. Br J Pharmacol 2007;152:1111–20.
 Rubovitch V, Gafni M, Sarne Y. The cannabinoid agonist DALN positively modulates L-type voltage-dependent calcium-channels in N18TG2 neuro- blastoma cells. Brain Res Mol Brain Res 2002;101:93–102.
 Fan SF, Yazulla S. Biphasic modulation of voltage-dependent currents of retinal cones by cannabinoid CB1 receptor agonist WIN 55212-2. Vis Neu- rosci 2003;20:177–88.
 Gonzalez B, Paz F, Floran L, Aceves J, Erlij D, Floran B. Cannabinoid agonists stimulate [3H]GABA release in the globus pallidus of the rat when G(i) protein-receptor coupling is restricted: role of dopamine D2 receptors. J Pharmacol Exp Ther 2009;328:822–8.
 Paquette JJ, Wang HY, Bakshi K, Olmstead MC. Cannabinoid-induced toler- ance is associated with a CB1 receptor G protein coupling switch that is prevented by ultra-low dose rimonabant. Behav Pharmacol 2007;18:767–76.
 Sanchez C, Rueda D, Segui B, Galve-Roperh I, Levade T, Guzman M. The CB(1) cannabinoid receptor of astrocytes is coupled to sphingomyelin hydrolysis through the adaptor protein fan. Mol Pharmacol 2001;59:955–9.
 Galve-Roperh I, Sanchez C, Cortes ML, Gomez dP, Izquierdo M, Guzman M.
Anti-tumoral action of cannabinoids: involvement of sustained ceramide accumulation and extracellular signal-regulated kinase activation. Nat Med 2000;6:313–9.
 Straiker AJ, Borden CR, Sullivan JM. G-protein alpha subunit isoforms couple differentially to receptors that mediate presynaptic inhibition at rat hippo- campal synapses. J Neurosci 2002;22:2460–8.
 Vasquez C, Lewis DL. The CB1 cannabinoid receptor can sequester G-proteins, making them unavailable to couple to other receptors. J Neurosci 1999;19:9271–80.
 Pacheco MA, Ward SJ, Childers SR. Identiﬁcation of cannabinoid receptors in cultures of rat cerebellar granule cells. Brain Res 1993;603:102–10.
 Shapira M, Vogel Z, Sarne Y. Opioid and cannabinoid receptors share a common pool of GTP-binding proteins in cotransfected cells, but not in cells which endogenously coexpress the receptors. Cell Mol Neurobiol 2000;20:291–304.
 Dainese E, Oddi S, Bari M, Maccarrone M. Modulation of the endocannabinoid system by lipid rafts. Curr Med Chem 2007;14:2702–15.